TSMC Approves Up to $1 Billion for Anthropic and OpenAI

Demand for AI chips supports Asian markets even as Wall Street trades unevenly. TSMC reports record October revenue of NT$586.89 billion, up 59.7% from a year earlier and 14.7% from September. Taiwan’s Taiex rises between 2.1% and 3.6% across the reports, while South Korea’s Kospi gains 1.6% to 2.4%, with Samsung Electronics and SK Hynix advancing. Hong Kong and Shanghai fall in some sessions, and China’s Singles’ Day shopping festival begins. TSMC forecasts fourth-quarter revenue of US$52.3 billion to US$54.3 billion and 2026 capital spending of US$60 billion to US$64 billion. Its board also approves up to US$1 billion of investment in Anthropic and OpenAI, which are developing their own processors; TSMC makes Apple’s custom chips. The articles also note Intel-related news preceded much of TSMC’s October share decline. Monthly sales do not reveal customer details or establish whether AI spending will continue at the same pace in 2027.

U.S. AI stocks rebound, with CoreWeave up 10.7% after posting a smaller-than-expected quarterly loss and raising its annual revenue forecast. CEO Michael Intrator says the company expects about US$9 billion in customer prepayments and deposits over the next three months, alongside other cash sources to cover bills. Micron Technology and Broadcom each gain 5.6%. Separately, Ramp’s data from more than 85,000 businesses shows October spending on OpenAI tokens rose 2% and Anthropic 1%, slower growth than in earlier months of 2026. Engineering teams account for 72% of AI spending, and typical firms spend about US$12 to US$15 per employee monthly. Ramp says SpaceXAI has the fastest adoption growth in October, with its share of businesses paying for the service up 2.2 percentage points to 6.3%; lower prices can also make spending totals diverge from usage growth.

Companies and institutions are also deciding how to deploy AI. Cisco, Goldman Sachs and JPMorgan are reviewing which tools suit different employee tasks, without specifying products or planned changes. At Robinhood, automated checks and an AI agent approve 60% of data-pipeline code changes without human review. Engineer Chetas Joshi reports a roughly 2% miss rate, 75% less manual review, about 132 engineering hours saved monthly and a 30% decline in related incidents. Sensitive changes, migrations that alter intent and low-confidence results still go to people, and teams continue weekly manual rule updates.

Texas 2036 reports that AI use in schools is outpacing policy: a fall 2025 survey found 86% of school and district leaders said teachers used AI and 61% reported student use, but only 29% had student-use policies and 20% had policies for teachers and staff. Texas is one of eight states without statewide AI guidance. The report urges statewide rules, protections for student safety and data, teacher training and instruction in responsible use, while noting the research on learning is early and shows potential benefits and harms depending on tools and guidance.

Research on AI oversight finds no finite, machine-checkable rule system can classify every AI-written program as safe or unsafe: some violations can be caught by observing a program, but finite tests cannot prove that it will never break a rule. Researchers say a gatekeeper can still produce a checkable list that includes programs it must allow and excludes those it must refuse, while leaving some cases undecided. Separately, Anthropic CEO Dario Amodei argues AI could transform treatment of cancer, Alzheimer’s, heart disease and mental illness. Scientists and drugmakers caution that AI may help identify candidates but cannot replace laboratory work, animal studies, human trials or regulatory review. Vertex Pharmaceuticals says none of about 50 AI firms it tested on 10,000 drug compounds produced a working drug on its own; as of 2026, no AI-designed drug had achieved full clinical adoption.

Decart releases Lucy 2.6, a live video model that the company says offers sharper images, more accurate details and improved lip syncing, with complex edits possible from one text prompt and a reference image. Its Delulu Stream app adds the new features, and Decart demonstrates live effects and scenery changes on mixed-reality jumbotron displays at a Colorado Avalanche game. In a separate sales-focused article, Patti Schutte of Be Brilliant Presentation Group says AI research can help teams prepare, but sellers should avoid reciting scripts. She recommends adapting to buyers’ responses, asking questions, exploring unexpected comments and noting after calls what worked and what to change.

Key Takeaways

  • TSMC reports record October revenue of NT$586.89 billion, up 59.7% year over year, and forecasts fourth-quarter revenue of US$52.3 billion to US$54.3 billion.
  • TSMC approves up to US$1 billion in investments in Anthropic and OpenAI; it also makes Apple’s custom chips, while OpenAI and Anthropic are developing processors of their own.
  • TSMC sets 2026 capital spending at US$60 billion to US$64 billion; its monthly sales figures do not identify customers or show whether AI spending will keep pace in 2027.
  • CoreWeave rises 10.7% after raising its annual revenue forecast; CEO Michael Intrator expects about US$9 billion in customer prepayments and deposits over the next three months.
  • Ramp finds October business spending on OpenAI tokens rose 2% and Anthropic 1%; engineering teams account for 72% of AI spending, while typical firms spend US$12–US$15 per employee monthly.
  • Robinhood’s hybrid system clears 60% of data-pipeline code changes without human review, saving about 132 engineering hours monthly; sensitive or uncertain changes still require people.
  • A Texas survey found teacher AI use at 86% of districts and student use at 61%, while only 29% had student-use policies and 20% had staff policies.
  • Researchers conclude finite rulebooks cannot certify every AI-written program as safe or unsafe, so a gatekeeper may need to leave some cases undecided.
  • Scientists caution that AI cannot replace drug testing and clinical review; Vertex says none of about 50 AI firms it tested on 10,000 compounds produced a working drug independently.
  • Decart’s Lucy 2.6 offers real-time video edits from a text prompt and reference image; sales adviser Patti Schutte urges teams to use AI for preparation but respond naturally to buyers.

Asian stocks rise as demand for AI chips lifts Taiwan and South Korea

Most Asian markets rose Wednesday, Nov. 11, as demand for AI chips helped lift shares in Taiwan and South Korea, despite a decline on Wall Street after U.S. inflation increased more than expected. Taiwan’s Taiex gained 2.1%, and Taiwan Semiconductor Manufacturing Co. rose 3.1% after reporting October revenue nearly six times higher than a year earlier. South Korea’s Kospi rose 1.6%, with Samsung Electronics up 2.3% and SK Hynix up 3.4%. Hong Kong and Shanghai markets fell, while China’s Singles’ Day shopping festival began and was expected to offer a sign of consumer demand.

Asian markets gain as AI chip demand offsets Wall Street losses

Asian shares were mostly higher Wednesday after U.S. inflation data pushed Wall Street lower. Taiwan’s Taiex rose 2.1%, while Taiwan Semiconductor Manufacturing Co. gained 3.1% after its October revenue came in nearly six times higher than a year earlier. South Korea’s Kospi advanced 1.6%, led by gains in Samsung Electronics and SK Hynix. Hong Kong’s Hang Seng and Shanghai’s Composite index fell, and China’s Singles’ Day shopping festival began, offering a measure of consumer demand.

Global stocks climb as chip demand lifts Asian markets

World markets were mostly higher Wednesday, with European indexes rising and Germany reporting that inflation fell to 2.8% in October. Taiwan’s Taiex gained 3.6%, and Taiwan Semiconductor Manufacturing Co. rose 3.1% after reporting October revenue nearly six times higher than a year earlier. TSMC’s board also approved up to $1 billion in investment in Anthropic and OpenAI. South Korea’s Kospi rose 2.4%, while Hong Kong’s Hang Seng fell 1.1%; China’s Singles Day shopping festival began, and India’s Sensex reached its highest level in six months. Wall Street had slipped Tuesday after U.S. inflation and bond yields rose, while the U.S. bond market was set to close Wednesday for Veterans Day.

AI stock rebound lifts Wall Street as CoreWeave raises its forecast

U.S. stocks rose Wednesday as shares in the artificial intelligence industry rebounded. CoreWeave jumped 10.7% after reporting a smaller quarterly loss than analysts expected and raising its revenue forecast for the year. CEO Michael Intrator said the company expects about $9 billion in customer prepayments and deposits over the next three months, along with other cash sources to pay its bills. Micron Technology and Broadcom each gained 5.6%, while Fossil Group rose 30.9% after reporting stronger profit and raising its revenue forecast. Trading was mixed overall, oil prices rose, and the U.S. bond market was closed for Veterans Day.

TSMC approves up to $1 billion for AI investments

Taiwan Semiconductor Manufacturing Co. approved up to $1 billion in investments in Anthropic and OpenAI, according to a company statement cited in the article. TSMC makes Apple’s custom chips, while OpenAI and Anthropic are developing processors of their own. The investments deepen TSMC’s ties with the two companies. The article was published Wednesday, Nov. 11, 2026.

TSMC sets October sales record as AI chip demand stays strong

TSMC reported October revenue of NT$586.89 billion, up 59.7% from a year earlier and 14.7% from September. The company said the figure was a record, and sales from January through October reached NT$4,485.62 billion, up 43.3% from the same period in 2025. TSMC shares fell about 11% in October, with most of the drop coming after news from Lam Research and Intel, not from TSMC’s sales report. The company expects fourth-quarter revenue of US$52.3 billion to US$54.3 billion and has set its 2026 capital spending budget at US$60 billion to US$64 billion. Monthly sales do not identify customers or show whether AI spending will continue at the same pace in 2027.

Patti Schutte says salespeople must turn AI research into real conversations

Patti Schutte, founder of Be Brilliant Presentation Group, says sales teams have more AI-generated research and talking points but still see deals stall. In recorded calls, she hears salespeople reciting scripts instead of responding naturally to buyers. Schutte advises teams to understand each deal well enough to ask questions, explore unexpected comments and adjust their approach. She helps one client replace a rigid script with a flexible story about how a buyer’s organization could change. After each call, she recommends writing down what worked, what was missed and what to do differently. Schutte says AI can help people prepare, but salespeople must use their own judgment to help buyers see the value of a solution.

Study finds finite rulebooks cannot certify every safe AI program

Researchers examine whether a gatekeeper can use fixed rules and machine-checkable evidence to decide which AI-written programs may run. They prove that two complete rulebooks, one for allowed programs and one for forbidden programs, cannot cover every case when evidence must be checked in finite time. A program’s violation of a prohibition can be caught by watching it, but no finite test can prove that it will never violate the rule. Rules about confidentiality and always responding also cannot be fully confirmed through finite evidence. The researchers say a gatekeeper can still produce a checkable list that includes everything it must allow and excludes everything it must refuse, without giving a verdict on every program. Tests involving 2.2 million random programs and self-referential adversaries found the problem was adversarial rather than typical.

Decart updates Lucy 2.6 for sharper real-time AI video edits

AI startup Decart released Lucy 2.6, an upgraded live video model available on its platform. Decart says its proprietary DOS infrastructure improves performance and stability, while the model produces sharper images, more accurate details and better lip syncing. Users can make more complex, multistep edits with a single text prompt and reference image. Decart encourages creators to use the model to change outfits, characters, products or settings during livestreams, and says it has added Lucy 2.6 features to its Delulu Stream app. The company also showed the model generating live effects and changing scenery on mixed reality jumbotron displays at a Colorado Avalanche hockey game.

Anthropic CEO predicts AI could transform medicine, but experts urge caution

Anthropic CEO Dario Amodei says AI could help make cancer, Alzheimer’s, heart disease and mental illness relics of the past. Other technology leaders, including Elon Musk and Google AI chief Demis Hassabis, have also made bold predictions about AI and medical breakthroughs. Drugmakers and scientists caution that AI may speed up the search for possible treatments but cannot remove lab testing, animal studies, human trials or regulatory reviews. Vertex Pharmaceuticals executive chairman Jeff Leiden said the company tested about 50 AI firms on 10,000 drug compounds, and none produced a working drug on its own. Critics also say current AI lacks the biological understanding needed to reliably create major cures, and no AI-designed drug had achieved full clinical adoption as of 2026.

Cisco, Goldman Sachs and JPMorgan rethink workplace AI access

Cisco, Goldman Sachs and JPMorgan are reconsidering which AI tools employees can use for different jobs. The companies still want workers to use AI, but they are reviewing which tools are best suited to specific tasks. The article does not name particular tools or explain what changes the companies may make.

Texas schools need clearer rules for AI use

A Texas 2036 report says AI use in schools is growing faster than the rules that guide it. In a fall 2025 survey, 86% of Texas school and district leaders said teachers used AI, and 61% reported student use, but only 29% had a policy for student AI use and 20% had one for teachers and staff. The report says research on AI and learning shows both benefits and harms, depending on the tools and how teachers guide their use. Texas is one of eight states without statewide AI guidance, while 42 states have issued some form of guidance. The report urges lawmakers to set statewide rules, protect student safety and data, train teachers, and teach students responsible AI use; it notes that the research remains early and has limits.

Robinhood AI system approves 60% of data code changes

Robinhood data platform engineer Chetas Joshi said a hybrid review system now merges 60% of data pipeline pull requests without human review. Scripts check data contracts, ownership and privacy tags, lineage, and freshness at production volume, while an AI agent examines code for risks the scripts may miss. In one example, the checks passed a change that widened a column from integer to bigint, but the agent found an unregistered reader that could cause a problem and sent it to a person for review. Joshi said the system’s miss rate was about 2%, manual review work fell about 75%, it saved about 132 engineering hours per month, and incidents tied to these pull requests fell about 30%. Sensitive data changes, migrations that change intent, and low-confidence results still require human review; weekly manual updates to the rules continue.

Ramp data shows AI spending growth slowed in October

Ramp’s AI Index found that business spending on OpenAI tokens rose 2% in October, while spending on Anthropic rose 1%. The index tracks payments from more than 85,000 U.S. businesses using Ramp’s corporate cards and bill-pay platform, and the October increases were much slower than growth in earlier months of 2026. Engineering teams made up 72% of AI spending, while typical firms spent about $12 to $15 per employee each month. Ramp economist Ara Kharazian said lower prices and changing competition can mean dollar spending does not match the growth in AI use. SpaceXAI had the fastest adoption growth among providers in October, with its share of businesses paying for the service rising 2.2 percentage points to 6.3%.

Sources

NOTE:

This news brief was generated using AI technology (including, but not limited to, Google Gemini API, Llama, Grok, and Mistral) from aggregated news articles, with minimal to no human editing/review. It is provided for informational purposes only and may contain inaccuracies or biases. This is not financial, investment, or professional advice. If you have any questions or concerns, please verify all information with the linked original articles in the Sources section below.

AI chips TSMC AI infrastructure semiconductor industry AI investment AI stocks CoreWeave AI business spending OpenAI Anthropic AI adoption AI coding agents AI in education AI policy AI safety verification AI drug discovery clinical trials generative video Lucy 2.6 AI in sales

Comments

Loading...